Excitons and their Fine Structure in Lead Halide Perovskite
Nanocrystals from Atomistic GW/BSE Calculations
Giulia Biffi,†,‡,kYeongsu Cho,¶,kRoman Krahne,†and Timothy C. Berkelbach∗,¶,§
†Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy
‡Dipartimento di Chimica e Chimica Industriale, Universit`a degli Studi di Genova, Via Dodecaneso, 31, 16146 Genova, Italy
¶Department of Chemistry, Columbia University, New York, New York 10027 USA
§Center for Computational Quantum Physics, Flatiron Institute, New York, New York 10010 USA
kThese authors contributed equally
E-mail: t.berkelbach@columbia.edu
Abstract
Atomistically detailed computational studies of nanocrystals, such
as those derived from the promising lead-halide perovskites, are
challenging due to the large number of atoms and lack of symme-
tries to exploit. Here, focusing on methylammonium lead iodide
nanocrystals, we combine a real-space tight binding model with the
GW approximation to the self-energy and obtain exciton wavefunc-
tions and absorption spectra via solutions of the associated Bethe-
Salpeter equation. We find that the size dependence of carrier con-
finement, dielectric contrast, electron-hole exchange, and exciton
binding energies has a strong impact on the lowest excitation en-
ergy, which can be tuned by almost 1 eV over the diameter range of
2–6 nm. Our calculated excitation energies are about 0.2 eV higher
than experimentally measured photoluminescence, and they display
the same qualitative size dependence. Focusing on the fine structure
of the band-edge excitons, we find that the lowest-lying exciton is
spectroscopically dark and about 20–30 meV lower in energy than
the higher-lying triplet of bright states, whose degeneracy is slightly
broken by crystal field effects.
Introduction
Lead-halide perovskite nanocrystals (LHP NCs), with formula
APbX3where A is a monovalent cation and X is a halide anion,
exhibit remarkable electronic and optical properties,1–3suggest-
ing promising applications in optoelectronics,4–8photonics,9,10 and
spintronics.11 Compared to bulk LHPs, LHP NCs display anoma-
lously short radiative lifetimes and concomitant high photolumines-
cence quantum yields, the origin of which is still under debate.1–3,12
These interesting optical properties and their temperature depen-
dence have focused attention on the exciton fine structure, i.e., the
energy ordering and character of the band-edge excitons.13?–19 For
example, the lowest-energy exciton of Cs-LHP NCs has been sug-
gested to be a bright (emissive) state with total angular momentum
J=1, which could explain the high photoluminescence quantum
yield.14–16 This property would be in stark contrast with the case
of conventional organic and inorganic semiconductors, where the
lowest-energy exciton is a dark (non-emissive) state. Other works
have suggested that this conventional behavior persists in the LHP
NCs (i.e., the lowest-energy exciton is a dark state with total an-
gular momentum J=0) but that exciton relaxation by phonons
is suppressed at low temperatures, yielding long lifetimes for the
bright exciton in both NCs17–19 and 2D LHPs.?
The properties of the band-edge excitons—whose energy sep-
arations are very small, of the order of 1–10 meV—are influ-
enced by quantum confinement, electron-hole attraction and ex-
change,14–16,20 dielectric environment,21 and lattice structure (in-
cluding the bulk and surface Rashba effects),22–25 making precise
theoretical predictions an incredible challenge. This complexity
has led to the development of exciton models based on effective-
mass or k·pHamiltonians, pioneered especially by Efros and co-
workers,14–16 among others.20,26 Atomistic simulation, although de-
sirable, is frustrated by the sizes of even the smallest experimentally
accessible NCs, which can have hundreds or thousands of atoms.
Although atomistic force-fields or density functional theory can be
applied to study the ground-state and structural properties of sys-
tems of this size,27–30 accurate excited-state theories are more ex-
pensive and generally inaccessible in a fully ab initio setting.
Here, we overcome this challenge and construct an atom-
istic orbital-dependent tight-binding model parametrized by first-
principles density functional theory (DFT) calculations, after which
we use a model dielectric function to apply self-energy corrections
via the GW approximation31–33 and calculate the energies and prop-
erties of excitons via the Bethe-Salpeter equation.34–36 The com-
putational approach is similar to previous work by two of us on
the properties of layered quasi-two-dimensional LHPs.37 As a spe-
cific example, we study methylammonium lead iodide (MAPbI3)
NCs, calculating the exciton binding energies, absorption spectra,
and band-edge fine structure. Within the approximations of our
approach, we find that the lowest-energy exciton is always a dark
state.
We study MAPbI3for several reasons, although our approach
is general and could be applied to any LHP NC. First, this LHP
is one of the most studied in its bulk form, especially for pho-
tovoltaics, partly due to its strong absorption at low energies that
facilitates sensitization.38,39 Moreover, bulk MAPbI3crystals have
low lasing thresholds40 and are generally more stable than Cs-based
ones.41 For these reasons, the atomic and electronic structure of
bulk MAPbI3has been extensively studied using DFT42,43 and the
GW approximation,44–47 providing important points of comparison.
These valuable properties have motivated experimental studies of
MAPbX3NCs with high photoluminescence quantum yields and
controllable size, leading to tunable band gaps. Such MAPbX3
NCs have been realized by colloidal synthesis48–53 and templated
growth inside porous oxide films,54,55 providing experimental re-
sults to which we can directly compare.
1
arXiv:2210.01324v1 [cond-mat.mtrl-sci] 4 Oct 2022